From a while now I’ve noticed that snapshots, don’t restore VMs to the state VMs were when one was taken.
I’ve been taken precautions to verify VM has no prior snapshots or that disks are consolidated (though most of them have been brand new VMs) and I power machines down (gracefully from the guest) before taking the snapshot, and when I try to go back (if the machine is running) when I restore the snapshot it will suddenly power off (to match the snapshot’s state when it was taken) but that’s about it. No newer data is deleted. It’s as if I hadn’t taken the snapshot at all.
Some weeks ago a VM had an issue where from the looks of it, its snapshots were messed up and vSphere started inflating each of its snapshots as if it were another copy of its disk and pretty soon I had a terabyte of written data for not even 70GB of real content. The VM could still turn on so I opted for was still accessible so I was able to rsync the data to network storage remove, delete and recreate the VM.
The strangest thing of all of this is that, normally bug-ridden, error-prone Windows, seems unaffected by it. I avoid snapshots at all costs but in Windows you just have to have ’em, and they’re working and snap back just fine.
I’m using vCenter 220.127.116.11000, ESXi from 13006603 to 14320388. I have not tried managing snapshots outside of vCenter. Are there known bugs on this version, was there something I maybe forgot to do before updating from an earlier one? I hope there’s a fix, it’s very expensive taking full clones as snapshots.